Académique Documents
Professionnel Documents
Culture Documents
Dans
les tunnels longs, la longueur du chemin de fuite dpend du type de
circulation et des possibilits quoffre le systme de ventilation. Mieux la
ventilation permet de rduire la zone o les chances de survie et les
conditions de visibilit sont limites, plus le trajet que devra parcourir un
usager pour svacuer (cest--dire linter-distance entre les issues de
secours) pourra tre long.
Dans les tunnels trs longs, o il est impossible de rduire la longueur de la
zone de danger par des mesures dexploitation, les issues de secours et les
itinraires dvacuation devront tre tudis trs srieusement. Les
possibilits suivantes existent :
5.3
le type de confinement,
lincendie et la fume auxquels il faut sattendre,
le comportement humain,
la ventilation,
la pente de la chausse (dclivit),
les contrles effectus dans le tunnel.
AIPCR
In short tunnels the portals are adequate. In longer tunnels, the length of the
escape route depends on the type of traffic and the ventilation system. The
more the limited survival zone and the bad visibility conditions can be
diminished by the ventilation system, the longer the escape distance for
tunnel users (i.e., interval between emergency routes) may be.
General recommendations for the distance between escape exits were not
given.
Montreal 1995
For the World Congress in Montreal, Working Groups 1 and 6 prepared a
paper on State of the Art: Existing Guidelines and Experiences on safety
equipment in tunnels within 13 countries, including emergency exits.
5.3
Traffic density is not taken into account as a factor to size the escape routes
because unless it is very low, the maximum number of road tunnel users
involved in an incident is determined by the length of the tunnel (i.e., the
number of cars stopped behind the fire) and not by the number of cars
passing per time increment. A typical risk analysis may conclude that escape
routes are not required for tunnels with very low traffic densities. In such a
case, it must be assessed whether the possibility of only a few people being
involved in a serious accident and the reduced probability of an accident can
justify the lack of escape routes.
PIARC
5.3.1
Types de confinement
Un ouvrage routier peut tre confin de la faon suivante :
5.3.2
AIPCR
With all structure types, it is assumed that without special provisions, there is
typically no chance for a road user to depart directly from the roadway. The
more enclosed a structure, the greater the effects of a fire inside the
structure. Since smoke rises during a fire, it can escape through gaps in the
ceiling. The larger the gaps, the slower the smoke will accumulate and
spread longitudinally along the tunnel. Depending on size and position
(height), this will hold true for gaps in sidewalls, too. Of course, provisions for
smoke extraction will help to limit the length of roadway being filled with
smoke. In general, the more a road structure is enclosed, the greater the
need for adequate escape provisions. This is demonstrated in Figure 5.1.
Toxic gases and vapours can be released during certain accidents, such as
when a tanker is damaged. Since the specific gravity of a gas or vapour is
typically higher than that of air, it will spread over the road surface whether or
not the structure has a roof. The extent of enclosure is therefore less
significant for toxic gas and vapour release than for a fire and smoke
situation, with the exception that side openings can help to dilute the gas
concentration if wind patterns are favourable. In tunnels with significant
longitudinal slopes, like underwater tunnels, toxic gases can cause severe
problems.
PIARC
Puissance de lincendie
On trouve dans la littrature existante de nombreuses valeurs pour la
puissance potentielle de lincendie. Le document AIPCR Matrise des
incendies et des fumes dans les tunnels routiers [2] donne les valeurs
suivantes :
2,5 MW
5 MW
8 MW
15 MW
10 MW
20-30 MW
30-250 MW
AIPCR
Fire Size
Many potential values for fire size can be found in existing literature. The
PIARC document Fire and Smoke Control in Road Tunnels [2] shows the
following values:
2.5 MW
5 MW
8 MW
15 MW
10 MW
20-30 MW
30-250 MW
PIARC
voiture
bus
petit camion avec chargement
grand camion avec chargement
camion-citerne
5-6 MW [22]
20-30 MW [23]
30-50 MW
50-150 MW [23]
200-300 MW [23]
AIPCR
Literature from the Netherlands provides the following maximum fire sizes:
car
bus
small loaded lorry
large loaded lorry
tanker
5-6 MW [22]
20-30 MW [23]
30-50 MW
50-150 MW [23]
200-300 MW [23]
Since the typical traffic mix is comprised mostly of cars, the probability of a
car fire is the highest. It is estimated that 95% of all car fires are caused by
breakdowns rather than accidents. Therefore, in the majority of incidents,
only one car is involved in the fire. A car fire is typically not very large, but
can still produce a lot of smoke. While not very high in temperature, the
smoke can be toxic enough to injure people. Very large fires that produce a
lot of heat and toxic smoke can occur when lorries are involved. An
investigation of the Elbe Tunnel in Germany from 1979 to 1985 indicated that
while lorries comprised only 15% of the total vehicle mix, 30% of all fires
were caused by lorries. Almost all of the fires in this investigation were
related to a vehicle breakdown.
If the cause of the fire is an accident, it often happens that more than one
vehicle is involved. In such events, the fire size is likely to be larger and
depends on the composition of the vehicles involved in the incident.
This is does not mean that the fire size will never be large in the case of a
vehicle breakdown, and conversely will never be small in the case of an
accident. The fire caused by a vehicle breakdown in the Mont Blanc Tunnel
was very large. The above information is only intended to give an idea of
what can reasonably be expected.
Fire Development
A vehicle fire is never immediately at its maximum size -- it takes time for the
fire to develop. In fact, almost all fires follow the same fire development
pattern, irrespective of the final fire size. An investigation including 30 car
fires in a closed parking facility [22] and a fire test in the Netherlands
Beneluxtunnel indicated the following typical scenario for a passenger car:
the fire remains small in the first 5-10 minutes after ignition;
the windows break, and the fire reaches its maximum size within 25 minutes;
PIARC
AIPCR
the fire burns at or near its maximum size for about 10-15 minutes;
Taking into account that most passenger car fires start in the engine, and that
passenger car engine compartments are similar to those of lorries and
busses, similar fire development patterns can be expected for larger fires.
Fire size and duration will, of course, be very different.
In the case of an accident, the fire will typically develop much faster due to
the release of gas and oil and the mixing of these combustibles with air. The
time for ignition is difficult to establish: it can occur very quickly after the
accident or some time later. It can be assumed that these fires will follow
similar initial development patterns as described above, regardless of final
fire size and duration.
Typical initial fire development is shown in Figure 5.2.
Figure 5.3 -
PIARC
5.3.3
Comportement humain
Un facteur important pour concevoir les dispositions dvacuation est le
comportement humain. Un rapport sur le comportement humain dans des
situations dangereuses [24 ; 25] indique que les ractions les plus courantes
vis--vis dun incendie sont lincrdulit, la sous-estimation et/ou le refus dun
danger potentiel. Ces sentiments se trouvent renforcs lors de la premire
phase dun incendie sil est faible et sans beaucoup de fume. Le rapport
conclut quenviron 37 % des personnes prsentes essaient dteindre le feu,
24 % avertissent les autres usagers, et seulement 16 % tentent de
schapper. Une grande partie des personnes qui restent sur place se
contentent de regarder lincendie, certaines courte distance.
Dautres personnes encore sont inconscientes de la gravit de la situation
parce quelles se trouvent une certaine distance de laccident, attendant
dans leurs voitures de pouvoir continuer. Elles peuvent tre alertes par
quelque signal durgence ou bien lorsquelles se trouvent recouvertes par la
fume, ce qui peut se rvler trop tard dans les deux cas. Lorsquun petit
incendie prend soudainement de lampleur, de nombreuses personnes se
trouvent mises en danger par les fumes et les gaz chauds.
Le comportement humain est tel que les gens essaient de faire, aussi
longtemps quils le peuvent, ce quils avaient lintention de faire ou ce quils
taient en train de faire avant de se trouver impliqus dans une situation
diffrente. En dautres termes, il est difficile de modifier leur tat desprit,
dune situation normale une situation requrant un comportement
inhabituel. Ce processus peut tre divis en trois phases (voir Figure 5.3) :
AIPCR
5.3.3
Human Behaviour
An important factor in designing escape provisions is human behaviour. A
report on human behaviour in dangerous situations [24; 25] indicates that the
most common reactions to fire are disbelief, underestimation and/or denial of
potential danger. These feelings are strengthened in the first phase of a fire
when it is small and without much smoke. The report concludes that about
37% of the people present will try to extinguish the fire, 24% will warn other
people, and only 16% will try to escape. Many people will just watch the fire,
from a short distance.
Other people will be unaware of the severity of the situation because they are
at some distance from the accident location, waiting in their cars until they
can drive on. They may be alerted by some kind of emergency call or when
they are covered by smoke, both of which may be too late. When a small fire
suddenly grows larger, many people are in danger because of the smoke and
heated gases.
Human behaviour is such that people try to do, as long as they can, what
they planned to do or what they were doing before they were involved in a
different situation. In other words, it is difficult to change their minds from a
normal situation to a situation requiring unusual behaviour. The process can
be divided into three phases (see Figure 5.3):
With regard to these three phases, the following comments are made:
the duration of the detection phase can vary significantly among different
road users, and greatly depends on the distance between the fire and the
observer. Lighting levels in tunnels are typically low, and visibility is
limited by the sidewalls and ceiling, so that smoke cannot be easily seen
from a distance;
PIARC
Le temps total dvacuation est la somme des dures des trois phases. Le
temps dvacuation disponible dpend de la vitesse laquelle lincendie se
dveloppe et de la propagation de la fume dans cette situation particulire.
Pour calculer le temps dvacuation, il faut connatre ou estimer au mieux la
dure des deux premires phases. Il existe peu de littrature disponible sur
ces trois phases car, pour autant que lon sache, ce sujet a fait lobjet de peu
de recherches. A partir de lexprience et de quelques exercices
dvacuation, il est possible de conclure que, sans alerte ou instructions
extrieures, les usagers ont besoin de 5 15 minutes pour dcider quils
doivent faire quelque chose. Si lon se base sur les dures normales de
dveloppement de lincendie mentionnes prcdemment, cela laisse aux
usagers trs peu de temps pour schapper, puisquils ont utilis la plus
grande partie du temps disponible pour prendre une dcision. Cette situation
est illustre par la figure 5.3.
Le retournement mental doit donc se faire aussi rapidement que possible,
pour que la phase dvacuation soit aussi longue que possible. Une dtection
rapide de lincendie et des instructions appropries donnes aux usagers du
tunnel peuvent raccourcir les deux premires phases, laissant ainsi plus de
temps pour lvacuation effective dans la troisime phase.
5.3.4
Ventilation
Un objectif important du systme de ventilation pendant un incendie est
dassurer que les voies dvacuation restent exemptes de fume. La
publication Matrise des incendies et des fumes dans les tunnels
routiers de lAIPCR, 1999 [2] donne des recommandations pour la
ventilation en cas dincendie. Celles-ci sont compltes par le chapitre 4 du
prsent rapport.
Les fumes menacent la vie des usagers pour plusieurs raisons. Une tude
faite sur des incendies dans des immeubles [26 ; 27] a montr quil est
impratif que les sorties de secours et les voies dvacuation restent visibles
dans les premiers instants de lincendie, o doit se raliser lauto-vacuation.
AIPCR
the decision made during the alarm phase determines the behaviour and
actions in the next phase. Because of the stress of the situation, it is very
difficult to change such decisions by external influences. This means that
a tunnel designer should provide as much information as possible to help
people make the right decisions. This information should also be more
noticeable (striking) than other visual or audible information;
the action phase includes the time to escape as well as to take other
actions.
The total evacuation time is the sum of the durations of the three phases.
The available evacuation time depends on the speed of fire development and
the spread of smoke in that particular situation.
In order to calculate the evacuation time, one should know, or make a best
guess at, the duration of the first two phases. There is limited literature
available on these phases because, as far as is known, limited research has
been conducted. From experience and some evacuation exercises, it can be
concluded that without external warning or instruction, people need about 515 minutes to decide that something has to be done. Based on the typical fire
development durations discussed above, this leaves people very little time to
actually escape, as they have used much of their available time waiting to
make a decision. This relationship is shown in Figure 5.3.
Mental turnaround must therefore take place as quickly as possible for the
effective evacuation phase to be as long as possible. Quick fire detection and
appropriate instructions to tunnel users will shorten the first two phases,
leaving more time for real escape in the third phase.
5.3.4
Ventilation
An important goal of the ventilation system during a fire is to provide smokefree escape routes. The publication Fire and Smoke Control in Road
Tunnels by PIARC, 1999 [2], provides recommendations for ventilation in
case of a fire. These are complemented by section 4 of the present report.
Smoke is life threatening to road users for several reasons. An investigation
of building fires [26; 27] showed that the visibility of emergency exits and
escape routes is imperative in the early stages of a fire, when self-evacuation
should take place.
PIARC
AIPCR
Tunnels shorter than 300 m are almost never equipped with mechanical
ventilation. Smoke movement is more difficult to predict when there is no
mechanical ventilation. The smoke can move around freely and is only
affected by natural air movement and buoyancy factors. As shown in
Figure 5.4, in tunnels with no mechanical ventilation the traffic space on one
or both sides of the fire can be affected by the smoke. If the escape route
from the traffic space is too long, then emergency exits should be provided.
As shown in Figure 5.5, partly closed structures are somewhat comparable to
short tunnels in that areas adjacent to roof openings are safe if all the
smoke can escape through the openings. The area between the roof
openings can become filled with smoke. If this area is too long, then
emergency exits should be provided.
In long tunnels, a distinction must be made between unidirectional and bidirectional traffic. As described in the PIARC report Fire and Smoke Control
in Road Tunnels [2], longitudinal ventilation is favoured over transverse
ventilation in unidirectional tunnels, road users are always present upstream
of the fire. A longitudinal ventilation system keeps the area upstream of the
fire smoke-free, which means that, in theory, there is no need for escape
routes. However, emergency exits should always be provided to account for
the unexpected, such as the fire developing to a size that the ventilation
system can no longer handle, or an explosion occurring.
PIARC
Figure 5.4 - Comportement de la fume dans un tunnel court sans ventilation mcanique
Smoke behaviour in a tunnel without mechanical ventilation
AIPCR
Figure 5.6 - Comportement de la fume dans un tunnel avec extraction des fumes
Smoke behaviour in a tunnel with smoke extraction
PIARC
5.3.5
AIPCR
5.3.5
The longitudinal slope will also influence the spread of burnable or burning
liquids, as well as the spread of smoke, since the tunnel ceiling slope is very
often the same as the road surface slope. In this case, escape provisions are
more influenced by the type and availability of ventilation.
PIARC
5.3.6
Contrles en tunnel
Dans un tunnel surveill 24 heures sur 24, il est habituellement possible de
dtecter rapidement un incendie (boucles dans la chausse, tlvision en
circuit ferm ou autres dispositifs), et des instructions peuvent normalement
tre donnes rapidement aux usagers (par haut-parleurs, transmission radio,
panneaux visuels, etc.), ce qui rend plus efficace lvacuation.
5.4
vacuation de secours
5.4.1
Issues de secours
Inter-distance
Les ttes dun tunnel constituent elles-mmes des issues de secours. Ainsi,
pour un tunnel ne possdant pas dissues de secours supplmentaires, la
longueur de litinraire dvacuation est gale la longueur totale du tunnel.
Ce sont le type douvrage et la longueur de litinraire dvacuation qui
dcideront de la ncessit ou non ditinraires dvacuation et de sorties de
secours supplmentaires.
Plus la distance entre les issues de secours est courte, plus le niveau de
scurit est lev. Il est relativement facile damnager des issues de
secours (passages transversaux) dun tube lautre si les tubes ne sont
spars que par une paroi ou un couloir dvacuation. Mais, dans le cas de
tunnels fors (en terrain rocheux ou meuble), lamnagement dissues de
secours peut devenir une tche important. Compte tenu de ce qui a t dit
dans les chapitres prcdents, on peut considrer ce qui suit :
AIPCR
5.3.6
Tunnel Controls
In a tunnel that is manned on a 24-hour basis, rapid fire detection is usually
possible (by loops in the road, CCTV or other provisions), and quick
instructions can usually be given to road users (by loud speakers, radio
broadcasting, visual signs, etc.), ensuring more efficient escape.
5.4
Emergency Egress
5.4.1
Emergency Exits
Spacing
The tunnel portals are themselves emergency exits. So, for a tunnel without
additional emergency exits, the length of the escape route is equal to the
total tunnel length. The type of structure and the length of the escape route
will dictate whether additional escape routes and emergency exits are
necessary.
The shorter the distance between emergency exits, the higher the safety
level. Providing emergency exits (cross passages) from tunnel tube to tunnel
tube is rather easy if the tubes are divided by only a wall or an escape
corridor. However, providing emergency exits for bored tunnels (in rock or
soft ground) can become a major task. Based on discussions in the previous
sections, the following should be considered:
a fire needs some time to develop, thus giving road users time to escape.
The development time is somewhat independent of the ultimate fire size;
road users usually hesitate to leave their cars and the potentially
dangerous area. Instructions by an operator can help them to escape
much quicker. Instructions can only be given if the tunnel is manned, if the
operators attention is drawn to the dangerous situation by some detection
system, and if the operator has the availability of communications
systems such as radio broadcasting, loudspeakers or visual signs;
PIARC